Virial statistical description of non-extensive hierarchical systems

In a first part the scope of classical thermodynamics and statistical mechanics is discussed in the broader context of formal dynamical systems, including computer programmes. In this context classical thermodynamics appears as a particular theory suited to a subset of all dynamical systems. A statistical mechanics similar to the one derived with the microcanonical ensemble emerges from dynamical systems provided it contains, 1) a finite non-integrable part of its phase space which is, 2) ergodic at a satisfactory degree after a finite time. The integrable part of phase space provides the constraints that shape the particular system macroscopical properties, and the chaotic part provides well behaved statistical properties over a relevant finite time. More generic semi-ergodic systems lead to intermittent behaviour, thus may be unsuited for a statistical description of steady states. Following these lines of thought, in a second part non-extensive hierarchical systems with statistical scale-invariance and power law interactions are explored. Only the virial constraint, consistent with their microdynamics, is included. No assumptions of classical thermodynamics are used, in particular extensivity and local homogeneity. In the limit of a large hierarchical range new constraints emerge in some conditions that depend on the interaction law range. In particular for the gravitational case, a velocity-site scaling relation is derived which is consistant with the ones empirically observed in the fractal interstellar medium.